A Parallelised High Performance Monte Carlo Simulation Approach for Complex Polymerisation Kinetics

Hugh Chaffey-Millar, Don Stewart, Manuel M. T. Chakravarty, Gabriele Keller, and Christopher Barner-Kowollik.

Macromolecular Theory & Simulation 16(6), pp 575-592, 2007.

A novel, parallelised approach to Monte Carlo simulations for the computation of full molecular weight distributions arising from complex polymerisation reactions is presented. The methods developed enable rapid simulation of the outcomes of reactions that previously required many hours or even days. Some systems (e.g. those arising in the authors' own research into star-polymer formation processes) which once required circa 8 hours of simulation time (using the commercial program package PREDICI) can now be simulated in a few minutes on a parallel computer.

In addition to this significant speed improvement, new insights have been developed with regard to the Monte Carlo process in at least four key areas: (1) an insufficient system size has been shown to create inaccuracies via poor representation of the most improbable events and least numerous species; (2) from a speed point of view, the method has been found to under-perform for some simple systems when compared to PREDICI but is vastly superior when applied to complex systems; (3) advanced algorithmic principles and compiler technology known to computer scientists have been used to provide speed improvements and (4) the parallelisability of the algorithm has been explored and excellent scalability been demonstrated.

For the first time, specific implementation strategies and their suitabilities are discussed. Further, the code is available, allowing other researchers to use or examine in detail the methods put forward.

Published version (from WILEY-VCH Verlag)
PDF (preprint) (37 pages)
supplementary data (PDF) (8 pages)

Project website with source code and a companion paper focusing on our contributions to programming languages.

This page is part of Manuel Chakravarty's WWW-stuff.